Pandas: how to add two dataframe with list element ignoring NaN?

Question

I want to use Pandas add operation to replace defaultdict(list) functional. If an index exists only in one dataframe, the result will be NaN. However, I hope the result is the e